Six AI pillars. One verified credential.
We don't compete on the 7-area AI list most platforms advertise. We built exactly six capabilities, each one named, each one exposed as a stable API, each one cost-routed so 85% of calls cost us (and you) ₹0.
The pipeline
Six pillars · one credential
Every submission flows through a deterministic 6-pillar pipeline. A hybrid inference stack keeps verification fast and scalable while evaluation-critical calls route to our frontier layer.
Playbook Forge
Employer brief → DIY learning playbook with injected tutorials.
In-house specialized inference
ExploreVerdict Engine
Our frontier evaluation layer scores work vs your rubric. Employer rates independently.
Frontier evaluation layer
ExploreIntegrity Sentinel
Plagiarism + AI-content + git-history + webcam heuristics on every submission.
In-house signal engine
ExploreRubric Blueprint
Unstructured brief → structured rubric with dimensions and weights.
In-house reasoning engine
ExploreProgress Navigator
Inline AI help scoped to the candidate’s current milestone — when they’re stuck.
In-house specialized inference
ExploreSkill Atlas
Maps submissions → standardised skill tags for portfolios, leaderboards, matching.
In-house signal engine
ExploreWhy 85/15 routing matters
Per Constitution §2.4, every AI call routes through our hybrid inference stack at 100.116.37.114:4000 running on a hybrid inference stack. Only calls that genuinely need frontier-model quality route to our frontier evaluation layer.
- • Target: 85% of AI calls served locally at ₹0
- • Claude: 15% of calls, for evaluation quality only
- • Per-candidate cost envelope: under ₹5
- • Monthly AI spend at 1,000 verified candidates/mo: under ₹5,000
Any code path that calls Anthropic directly without documented rationale fails CI. Cost discipline is a constitutional requirement, not a cost-optimisation hobby.